The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to install required Python dependencies from a local requirements.txt file and to download spaCy language models. This is a standard and expected configuration for NLP-based tools.
[EXTERNAL_DOWNLOADS]: The instruction set includes commands to fetch English and French language models using the official spaCy download utility. These downloads are directed to well-known registries associated with the library's official maintainers.
[DATA_EXFILTRATION]: While the skill creates audit logs containing entity mappings for reversible anonymization strategies, it does not perform any network exfiltration. The skill documentation includes explicit safety rules for the user, such as securing the generated audit logs and adding them to .gitignore to prevent accidental exposure via version control.
[PROMPT_INJECTION]: The skill processes untrusted user-supplied documents for entity detection. This presents a surface for indirect prompt injection where an attacker might attempt to influence the agent's behavior through instructions embedded in the document text.
Ingestion points: Files processed via detect.py and anonymize.py (SKILL.md).
Boundary markers: None identified; the scripts read the raw text contents of the target files.
Capability inventory: The agent has access to Bash, Write, and Read tools.
Sanitization: The skill uses regex and NER for detection but does not perform content sanitization specifically designed to strip prompt injection payloads.

anonymize-doc